Pre-coding and post-processing used by a transmission device of a multi-antenna multi-user communication system are classified into liner types and nonlinear types. The linear pre-coding can be subdivided into unitary pre-coding and non-unitary pre-coding. Coordinated beamforming belongs to non-unitary pre-coding.
In coordinated beamforming, a transmission device calculates pre-coding matrices and receive beamforming vectors (or receive beamforming matrices) using downlink channel information to all active users. The pre-coding matrices and the receive beamforming vectors are calculated using an iteration algorithm so that inter-user interference may be minimized. The inter-user interference is also known as multiple-access interference.
Coordinated beamforming uses the following two methods to calculate transmit/receive beamforming vectors.
In a first method, using pilot beamforming, a transmission device allocates dedicated pilots to respective reception devices. Thereafter, the transmission device beam-forms dedicated pilots as receive beamforming vectors for reception devices and transmits them. Each of the reception devices estimates an effective channel using the dedicated pilot, forms a matched filter on the estimated effective channel, and uses it as a receive beamforming vector.
In a second method, a transmission device quantizes receive beamforming vectors for respective reception devices and transmits them to the reception devices over a feedforward channel. One of the two methods is selected depending on whether or not the multiantenna multiuser communication system uses dedicated pilots.
Meanwhile, since coordinated beamforming uses an iteration algorithm that converges pre-coding vectors and receive beamforming vectors on a value at which user interference is minimized, its calculation is given as a random variable. Therefore, signal transmission by the transmission device may be delayed until calculation of transmit/receive beamforming vectors has converged. On the other hand, when iteration of the iteration algorithm is restricted by compulsion, the transmit/receive beamforming vectors may be calculated incorrectly, causing a reduction in performance.
In addition, after calculating the transmit/receive beamforming vectors using the iteration algorithm, coordinated beamforming must go through two steps of quantizing receive beamforming vectors using a codebook, and readjusting the transmit beamforming vectors using the quantized receive beamforming vectors. Therefore, in some cases performance may be degraded due to the quantization.